Unconstrained face verification using deep CNN features

By Jun-Cheng Chen, Vishal M. Patel, Ramalingam Chellappa. IEEE Workshop/Winter Conference on Applications of Computer Vision, 2015

302 citations recorded by the source corpus.

In this paper, we present an algorithm for unconstrained face verification based on deep convolutional features and evaluate it on the newly released IARPA Janus Benchmark A (IJB-A) dataset as well as on the traditional Labeled Face in the Wild (LFW) dataset. The IJB-A dataset includes real-world unconstrained faces from 500 subjects with full pose and illumination variations which are much harder than the LFW and Youtube Face (YTF) datasets. The deep convolutional neural network (DCNN) is trained using the CASIA-WebFace dataset. Results of experimental evaluations on the IJB-A and the LFW datasets are provided.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

View Paper: https://www.semanticscholar.org/paper/99b2d0c2fa7c3155b0a8f1e283ea5d6165919307

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Signal Processing

Citations in the corpus

Towards Large-Pose Face Frontalization in the Wild (2017, methodology): “Some other methods focus on fusing information at the feature level [2] or distance level [16].”
Are Accuracy and Robustness Correlated (2016, background): “Due to the significant performance improvements on visual recognition tasks, stateof-the-art machine learning models have managed to obtain classification accuracies comparable to or even better than human-level performance [1], [2], [3], [4].”
Face recognition accuracy of forensic examiners, superrecognizers, and face recognition algorithms (2018, unknown): “Fusing individual examiners with A2017a and A2016 yielded performance equivalent to the fusion of two examiners (Mann–Whitney U test = 4.53× 104, n1 = 1,596, n2 =57, P=0.956; Mann–Whitney U test = 4.33× 104, n1 =1,596, n2 =57, P=0.526, respectively).”
Person Recognition in Personal Photo Collections (2017, background): “[6], [7], [8], [9], [10], [11], [12], [13] trained on large scale face”
Deep Face Recognition: A Survey (2018, methodology): “[96], [103], [81] which first aggregates face representations into a single representation for each set and then performs a comparison between two sets.”
LOTS about attacking deep features (2016, methodology): “To form a more robust representation of an identity, deep features of several images are averaged [1].”
Attention-Aware Deep Reinforcement Learning for Video Face Recognition (2017, background): “Video face recognition has attracted great attention in computer vision over the past few years [4, 7, 8, 15, 24, 31, 32, 40, 41, 43].”
Racial Faces in the Wild: Reducing Racial Bias by Information Maximization Adaptation Network (2018, unknown): “38 Joint Bayesian [17] - 83.”
Learning Pose-Aware Models for Pose-Invariant Face Recognition in the Wild (2019, methodology): “[14], [31] demonstrated compelling results on IJB-A by using a single CNN trained from scratch on frontal views and”
L2-constrained Softmax Loss for Discriminative Face Verification (2017, methodology): “Most of the recent methods [29, 30, 24, 37] train a DCNN with softmax loss to learn these features which are later used either to directly compute the similarity score for a pair of faces or to train a discriminative metric embedding [27, 3].”
FaceID-GAN: Learning a Symmetry Three-Player GAN for Identity-Preserving Face Synthesis (2018, unknown): “2 DCNN [5] 78.”
Do We Really Need to Collect Millions of Faces for Effective Face Recognition? (2016, methodology): “4 The results reported in [4] with fine tuning on the training sets include system components not evaluated without fine tuning.”
Frontal to profile face verification in the wild (2016, methodology): “In training the deep network [9], the author used 7 key-points on both sides of the face to align the images.”